Hands-On Machine Learning for Algorithmic Trading

Predictive performance by factor quantiles

As a first step, we would like to visualize the average period return by factor quantile. We can use the built-in function mean_return_by_quantile from the performance and plot_quantile_returns_bar from the plotting modules:

from alphalens.performance import mean_return_by_quantilefrom alphalens.plotting import plot_quantile_returns_barmean_return_by_q, std_err = mean_return_by_quantile(alphalens_data)plot_quantile_returns_bar(mean_return_by_q);

The result is a bar chart that breaks down the mean of the forward returns for the four different holding periods based on the quintile of the factor signal. As you can see, the bottom quintiles yielded markedly more negative results than the top quintiles, ...